Abstract
Progeny test evaluations for milk yield by modified contemporary comparison of 12,207 Holstein bulls (sons) were predicted from genetic evaluations of their sires, dams and maternal grandsires. Evaluations of sires and maternal grandsires were contemporary comparisons of records of daughters initiated prior to the birth of each son. Evaluations of dams were from 1st, later and all available records initiated prior to birth of sons. Sons, sires and maternal grandsires were classified as in artificial insemination or not. Intensity of selection by pedigree was higher for sons in artificial insemination. Regressions of evaluations of sons on those of ancestors generally were less in bulls in artificial insemination than for other bulls. Intercepts to predict evaluations of sons varied widely across populations of bulls, suggesting improvements are possible in adjustments of evaluations for genetic improvement. Evaluations of dams by production in 1st lactations were most useful in predicting evaluations of sons. Adding evaluations of dams to those for sires and maternal grandsires did not increase the accuracy of prediction of bulls in artificial insemination.